As integrated circuit transistor densities increase, and feature sizes shrink, capacitive coupling between adjacent interconnects, metal lines or other elements also increases. The increased capacitive coupling results in increased parasitic capacitance, which undesirably slows circuit speeds and negatively impacts overall device performance.
Current attempts to improve electrical isolation in high density integrated circuits involve the implementation of low K dielectric materials such as hydrogen silsesquioxane (HSQ), SiLK™ (a trademark of The Dow Chemical Company) resin, Black Diamond™ (a trademark of Applied Materials company) low K film, Coral™ (a trademark of Novellus System Inc.) carbonaceous oxide film and several other exotic materials. While these materials have a relatively low dielectric constant, they are not normally used in semiconductor manufacturing and therefore increase manufacturing complexity and costs. Much work remains to effectively integrate these materials into conventional semiconductor manufacturing processes.
Some disadvantages of current low K materials include incompatible thermal coefficient of expansion, low mechanical strength and poor thermal diffusivity.
Another manner of improving electrical isolation between interconnects is to use an integrated air gap structure because of the extremely low dielectric constant of air. Previous attempts at air gap structures were hard to manufacture and also did not completely isolate adjacent metal lines due to fringing fields above and below the air gap itself.
For example, U.S. Pat. No. 6,177,329 to Pang (and particularly at col. 7, 11. 46+) illustrates one conventional approach in which an additional mask is used to pattern the underlying layers to form the air gaps. This is both inefficient and imprecise for extremely small geometries. U.S. Pat. No. 5,847,439 to Reinberg illustrates another approach in which a combination of a low melting point dielectric, photoresist, a heat cycle and surface tension interact to form a void between two adjacent metal lines. This technique is clearly not suitable for precise control of air gap sizes, and is further disadvantageous because it cannot be used to form gaps which extend above a metal line. The latter may be desirable in some applications. Finally, U.S. Pat. No. 5,949,143 to Bang depicts a rather complex process in which a small opening is made in an etch stop layer and then a selective isotropic etch is used to remove dielectric between two metal lines.
Clearly, while portions of the aforementioned references are useful in forming air gap structures, and could be used in many applications, their overall approach is not optimal from a manufacturing perspective.
What is desired, therefore, is an easily manufacturable integrated air gap structure that substantially electrically isolates adjacent interconnects, metal lines or other IC elements.